Data quality is not one property, so asking whether data is good has no single answer.
The dimensions split the question into parts that can each be measured and owned.
Each one asks something different of the same table.
Accuracy asks whether values match the real world, such as a price matching its invoice.
Completeness asks whether the values that should exist do exist, such as every order having a price.
Consistency asks whether the same fact agrees across places, such as one customer's address in 2 systems.
Validity asks whether values follow the rules for their field, such as a date being a real calendar date.
Uniqueness asks whether each real thing appears once, so one customer does not become 2.
Timeliness asks whether the data is recent enough for its use, such as stock levels updated within the hour.
A value can be valid but inaccurate, like a correctly formatted phone number that belongs to someone else.
Accuracy is the hardest to measure, because it needs a trusted source to compare against.
The other dimensions can usually be checked by code alone, which is why they are checked far more often.
Teams often track a score per dimension, such as the share of rows with a valid email, and watch it over time.
Which dimensions matter most depends on the use: a monthly report can tolerate a delay that a fraud check cannot.
The common misconception is that tidy-looking data is accurate, when format checks say nothing about truth.